khadhroony-bot3/scripts/audit_khadhroony_workspace_rules.py

#!/usr/bin/env python3
# file: scripts/audit_khadhroony_workspace_rules.py
# version: 2

"""Audit mechanically verifiable rules specific to khadhroony-bot3."""

from __future__ import annotations

import argparse
import dataclasses
import pathlib
import re
import sys
import tomllib


@dataclasses.dataclass(frozen=True)
class Violation:
    """One project-specific workspace rule violation."""

    code: str
    path: str
    line: int
    message: str


def package_name(crate_dir: pathlib.Path) -> str | None:
    """Return a crate package name."""

    cargo = crate_dir / "Cargo.toml"
    if not cargo.exists():
        return None
    value = tomllib.loads(cargo.read_text(encoding="utf-8")).get("package", {}).get("name")
    return value if isinstance(value, str) else None


def relevant_tracing_crate(crate_dir: pathlib.Path) -> bool:
    """Return whether the crate owns a tracing target contract."""

    cargo = crate_dir / "Cargo.toml"
    data = tomllib.loads(cargo.read_text(encoding="utf-8"))
    if "tracing" in data.get("dependencies", {}):
        return True
    return any("TRACING_TARGET" in path.read_text(encoding="utf-8") for path in crate_dir.glob("src/**/*.rs"))


def audit_tracing(root: pathlib.Path) -> list[Violation]:
    """Audit the project tracing target contract."""

    violations: list[Violation] = []
    for cargo in sorted(root.glob("*/Cargo.toml")):
        crate = cargo.parent
        if crate.name == "kb-lib":
            continue
        if not relevant_tracing_crate(crate):
            continue
        name = package_name(crate)
        if name is None:
            continue
        constants = crate / "src/constants.rs"
        expected = f'pub(crate) const TRACING_TARGET: &str = "{name}";'
        relative_constants = constants.relative_to(root).as_posix()
        if not constants.exists() or expected not in constants.read_text(encoding="utf-8"):
            violations.append(Violation("KH_TRACE001", relative_constants, 1, f"expected `{expected}`"))
        crate_root = crate / "src/lib.rs"
        if not crate_root.exists():
            crate_root = crate / "src/main.rs"
        relative_root = crate_root.relative_to(root).as_posix()
        if not crate_root.exists() or "pub(crate) use crate::constants::TRACING_TARGET;" not in crate_root.read_text(encoding="utf-8"):
            violations.append(Violation("KH_TRACE002", relative_root, 1, "missing crate-root TRACING_TARGET re-export"))
        for path in crate.glob("src/**/*.rs"):
            relative = path.relative_to(root).as_posix()
            for index, line in enumerate(path.read_text(encoding="utf-8").splitlines(), 1):
                if "khbot." in line:
                    violations.append(Violation("KH_TRACE003", relative, index, "legacy khbot tracing target is forbidden"))
                if "crate::constants::TRACING_TARGET" in line and path.name not in {"lib.rs", "main.rs"}:
                    violations.append(Violation("KH_TRACE004", relative, index, "use crate::TRACING_TARGET outside crate root"))
                if re.search(r"(?:crate::[A-Za-z0-9_]+::|(?<!crate::))TRACING_TARGET", line) and "const TRACING_TARGET" not in line and "use crate::constants::TRACING_TARGET" not in line:
                    if "crate::TRACING_TARGET" not in line:
                        violations.append(Violation("KH_TRACE005", relative, index, "non-canonical TRACING_TARGET path"))
    return violations


def audit_kb_lib_tracing(root: pathlib.Path) -> list[Violation]:
    """Audit hierarchical tracing targets owned by operational kb-lib components."""

    violations: list[Violation] = []
    crate = root / "kb-lib"
    crate_root = crate / "src/lib.rs"
    if not crate_root.exists():
        return violations
    root_text = crate_root.read_text(encoding="utf-8")
    aliases = set(
        re.findall(
            r"^pub\(crate\) use crate::[A-Za-z0-9_:]+::([A-Z][A-Z0-9_]+_TRACING_TARGET);$",
            root_text,
            re.MULTILINE,
        )
    )
    target_pattern = re.compile(
        r'^pub\(crate\) const TRACING_TARGET: &str = "([^"]+)";$',
        re.MULTILINE,
    )
    for constants in sorted((crate / "src").rglob("constants.rs")):
        text = constants.read_text(encoding="utf-8")
        match = target_pattern.search(text)
        if match is None:
            continue
        component = constants.parent.relative_to(crate / "src")
        expected = "kb-lib." + ".".join(component.parts)
        relative = constants.relative_to(root).as_posix()
        if match.group(1) != expected:
            violations.append(
                Violation(
                    "KH_TRACE101",
                    relative,
                    text[: match.start()].count("\n") + 1,
                    f"expected hierarchical target `{expected}`",
                )
            )
    macro_pattern = re.compile(
        r"tracing::(?:debug|error|info|trace|warn)!\(\s*target:\s*crate::([A-Z][A-Z0-9_]+_TRACING_TARGET)"
    )
    for path in sorted((crate / "src").rglob("*.rs")):
        relative = path.relative_to(root).as_posix()
        text = path.read_text(encoding="utf-8")
        for index, line in enumerate(text.splitlines(), 1):
            if "let _target" in line:
                violations.append(
                    Violation(
                        "KH_TRACE102",
                        relative,
                        index,
                        "fake tracing target consumption is forbidden",
                    )
                )
        for match in macro_pattern.finditer(text):
            if match.group(1) not in aliases:
                violations.append(
                    Violation(
                        "KH_TRACE103",
                        relative,
                        text[: match.start()].count("\n") + 1,
                        f"missing crate-root re-export for `{match.group(1)}`",
                    )
                )
    return violations


def audit_solana_types(root: pathlib.Path) -> list[Violation]:
    """Reject non-canonical Solana address types in workspace Rust code."""

    violations: list[Violation] = []
    for path in sorted(root.rglob("*.rs")):
        if "target" in path.parts or ".git" in path.parts:
            continue
        relative = path.relative_to(root).as_posix()
        for index, line in enumerate(path.read_text(encoding="utf-8").splitlines(), 1):
            if "solana_address::Address" in line:
                violations.append(Violation("KH_SOL001", relative, index, "use solana_pubkey::Pubkey instead of solana_address::Address"))
    return violations


def main() -> int:
    """Run the khadhroony-specific audit."""

    parser = argparse.ArgumentParser()
    parser.add_argument("--root", default=".")
    parser.add_argument("--report-only", action="store_true")
    arguments = parser.parse_args()
    root = pathlib.Path(arguments.root).resolve()
    violations = audit_tracing(root) + audit_kb_lib_tracing(root) + audit_solana_types(root)
    violations.sort(key=lambda item: (item.code, item.path, item.line, item.message))
    if not violations:
        sys.stdout.write("Khadhroony workspace rule audit: clean\n")
        return 0
    sys.stdout.write(f"Khadhroony workspace rule audit: {len(violations)} violation(s)\n")
    for item in violations:
        sys.stdout.write(f"{item.code} {item.path}:{item.line}: {item.message}\n")
    return 0 if arguments.report_only else 1


if __name__ == "__main__":
    raise SystemExit(main())